Thursday, December 16, 2010

Predicting Sustainable Retirement Withdrawal Rates Using Valuation and Yield Measures

Update on May 16, 2011: I have substantially re-written this article. A description of the revised article can be found here.

In my last post, I mentioned hoping to finish two more articles before the holidays.  Well, one of those articles will have to wait until January.  But one of them is finished now!

This article is "Predicting Sustainable Retirement Withdrawal Rates Using Valuation and Yield Measures."  The full article can be downloaded from RePEc  or SSRN.

This article basically represents the third part of a trilogy of articles that each uses an almost completely different methodology to conclude that U.S. retirees in the past decade may experience the worst retirement outcomes in history.  This article is now the most specific, as I am finally estimating the sustainable withdrawal rates for recent retirees. Here is the abstract:

This study attempts to quantify whether a 4 percent withdrawal rate can still be considered as safe for U.S. retirees in recent years when earnings valuations have been at historical highs and the dividend yield has been at historical lows. We find that the traditional 4 percent withdrawal rule is likely to fail for recent retirees. The maximum sustainable withdrawal rate (MWR) for retirees may continue declining even after the peak in earnings valuations in 2000. Our lowest point estimate for an MWR with a 60/40 allocation between stocks and bonds is 1.46 percent for new retirees in 2008. We also discuss confidence intervals for these predictions. The regression framework with variables to predict long-term stock returns, bond returns, and inflation (the components driving the retiree's remaining portfolio balance) produces estimates that fit the historical data quite well, and we use backtesting for a further robustness check. Nevertheless, there are important qualifications for these predictions. In particular, they depend on out-of-sample estimates as the circumstances of the past 15 years have not been witnessed before, and there is always potential for structural changes which could leave recent retirees in better shape than suggested by the model. Looking forward, this methodology can guide new retirees toward a reasonable range for their MWR so that the 4 percent rule need not be blindly followed.

And finally, here is a figure which tells the basic story: the regression model fit the historical data quite well, and its predictions since 1980 (the last year that we actually know the 30-year sustainable withdrawal rate) shows the bad news, this case for a 60/40 stock and bond asset allocation:

Happy Holidays!